[Firm] A question about spilling in register allocation

Sebastian Hack hack at cs.uni-saarland.de
Mon Jun 21 15:42:11 CEST 2021


indeed, the algorithm is very similar. The message / contribution is: If you apply a simple algorithm as this on an SSA-form program, you get an optimal coloring of the interference graph. In non-SSA settings this is not the case.


> On 21. Jun 2021, at 15:36, 五月的风 <anlarry at foxmail.com> wrote:
> Hello, Firm developers: 
> I'm a student learning compiler design. I read your wonderful book, Register Allocation for Programs in SSA form. I'm comfused that the algorithm in section 4.2.4, it's just like linear scan for spilling and coloring. I wonder what the difference between linear scan and the algorithm proposed in the book.
> Thanks!
> _______________________________________________
> Firm mailing list
> Firm at ira.uni-karlsruhe.de
> https://lists.ira.uni-karlsruhe.de/mailman/listinfo/firm

More information about the Firm mailing list